NAEP Technical DocumentationSampling Weights and Disproportionate Representation in the 2000 National Main Assessment

The national main assessment uses differential sampling rates. Differential sampling rates oversample specific student groups to obtain larger samples of respondents from these groups. This oversampling enhances the precision of estimates of characteristics of these groups.

For the 2000 national main assessment, oversampling was achieved at either the school or student sampling stages and, in the case of Black and Hispanic students, at both stages. Oversampled student groups included:

students with disabilities or limited-English-proficient (SD/LEP) students,
Black students,
Hispanic students, and
private school students.

Very small schools (defined as having 1 to 19 eligible students) are sampled at a lower rate than other schools. The lower level of sampling for small schools reduces variances per unit of cost since it is relatively costly to administer assessments in small schools. Appropriate estimation of population characteristics takes disproportionate representation into account. A weight is assigned to each respondent, where the weight appropriately accounts for the sample design and reflects the proportional representation of the various types of individuals in the population.

The weighting procedures involve the calculation of final student weights for various student samples and reporting populations. A final student weight is the product of several weighting components including school and student base weights and various weighting adjustments at both the school and student level to improve precision.

Last updated 03 March 2008 (PE)

Printer-friendly Version